A client that has fallen back could detect an inappropriate fallback if
the TLSv1.3 downgrade protection sentinels are present.
Fixes#6756
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6894)
At certain points in the handshake we could receive either a plaintext or
an encrypted alert from the client. We should tolerate both where
appropriate.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6887)
If we sent early_data and then received back an HRR, the enc_write_ctx
was stale resulting in errors if an alert needed to be sent.
Thanks to Quarkslab for reporting this.
In any case it makes little sense to encrypt alerts using the
client_early_traffic_secret, so we add special handling for alerts sent
after early_data. All such alerts are sent in plaintext.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6887)
Under certain error conditions a call to SSLfatal could accidently be
missed.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6872)
In some scenarios the connection could fail without an alert being sent.
This causes a later assertion failure.
Thanks to Quarkslab for reporting this.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/6852)
In particular, adhere to the rule that we must not modify any
property of an SSL_SESSION object once it is (or might be) in
a session cache. Such modifications are thread-unsafe and have
been observed to cause crashes at runtime.
To effect this change, standardize on the property that
SSL_SESSION->ext.hostname is set only when that SNI value
has been negotiated by both parties for use with that session.
For session resumption this is trivially the case, so only new
handshakes are affected.
On the client, the new semantics are that the SSL->ext.hostname is
for storing the value configured by the caller, and this value is
used when constructing the ClientHello. On the server, SSL->ext.hostname
is used to hold the value received from the client. Only if the
SNI negotiation is successful will the hostname be stored into the
session object; the server can do this after it sends the ServerHello,
and the client after it has received and processed the ServerHello.
This obviates the need to remove the hostname from the session object
in case of failed negotiation (a change that was introduced in commit
9fb6cb810b in order to allow TLS 1.3
early data when SNI was present in the ClientHello but not the session
being resumed), which was modifying cached sessions in certain cases.
(In TLS 1.3 we always produce a new SSL_SESSION object for new
connections, even in the case of resumption, so no TLS 1.3 handshakes
were affected.)
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6378)
These tiny functions only read from the input SSL, and we are
about to use them from functions that only have a const SSL* available,
so propagate const a bit further.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6378)
The spec says that a client MUST set legacy_version to TLSv1.2, and
requires servers to verify that it isn't SSLv3.
Fixes#6600
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6747)
If we issue new tickets due to post-handshake authentication there is no
reason to remove previous tickets from the cache. The code that did that
only removed the last session anyway - so if more than one ticket got
issued then those other tickets are still valid.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6722)
Check that we are either configured for PSK, or that we have a TLSv1.3
capable certificate type. DSA certs can't be used in TLSv1.3 and we
don't (currently) allow GOST ones either (owing to the lack of standard
sig algs).
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6650)
In TLSv1.2 and below we should remove an old session from the client
session cache in the event that we receive a new session ticket from the
server.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/6601)
During anti-replay we cache the ticket anyway, so there is no point in
using a full stateless ticket.
Fixes#6391
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6563)
Implement support for stateful TLSv1.3 tickets, and use them if
SSL_OP_NO_TICKET is set.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6563)
Separate out as a new function the code to write out data which is specific
to a stateless ticket.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6563)
Sessions should be immutable once they are in the cache because they could
be shared with other threads. If you change them then this can cause
corruptions and races
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6561)
TLSv1.3 ignores the alert level, so we should suppress sending of
warning only alerts.
Fixes#6211
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6370)
We generate the secrets based on the nonce immediately so there is no
need to keep the nonce.
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6415)
All tickets on a connection need to have a unique nonce. When this was
originally implemented we only ever sent one ticket on the conneciton so
this didn't matter. We were just using the value 0. Now we can get multiple
tickets to we need to start doing the ticket nonce properly.
Fixes#6387
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6415)
We send a session ticket automatically in TLSv1.3 at the end of the
handshake. This commit provides the ability to set how many tickets should
be sent. By default this is one.
Fixes#4978
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5227)
In DTLS if we have buffered a fragment for a zero length message (e.g.
ServerHelloDone) then, when we unbuffered the fragment, we were attempting
to memcpy the contents of the fragment which is zero length and a NULL
pointer. This is undefined behaviour. We should check first whether we
have a zero length fragment.
Fixes a travis issue.
[extended tests]
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6223)
Don't call the decrypt ticket callback if we've already encountered a
fatal error. Do call it if we have an empty ticket present.
Change the return code to have 5 distinct returns codes and separate it
from the input status value.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6198)
The return value from the ticket_key callback was not properly handled in
TLSv1.3, so that a ticket was *always* renewed even if the callback
requested that it should not be.
Also the ticket decrypt callback was not being called at all in TLSv1.3.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6198)
During a full handshake the server is the last one to "speak". The timer
should continue to run until we know that the client has received our last
flight (e.g. because we receive some application data).
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6170)
In 1.0.2 and below we always send the same client_version in a reneg
ClientHello that we sent the first time around, regardless of what
version eventually gets negotiated. According to a comment in
statem_clnt.c this is a workaround for some buggy servers that choked if
we changed the version used in the RSA encrypted premaster secret.
In 1.1.0+ this behaviour no longer occurs. This restores the original
behaviour.
Fixes#1651
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6059)
This dead code should have been removed as part of #5874 but got missed.
Found by Coverity.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/6049)
In 1.1.0 and before calling SSL_in_init() from the info_callback
at SSL_CB_HANDSHAKE_DONE would return 0. This commit fixes it so
that it does again for 1.1.1. This broke Node.
Fixes#4574
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6019)
The first session ticket sent by the server is actually tacked onto the
end of the first handshake from a state machine perspective. However in
reality this is a post-handshake message, and should be preceeded by a
handshake start event from an info callback perspective.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5874)
We cannot provide a certificate status on a resumption so we should
ignore this extension in that case.
Fixes#1662
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/5896)
We have been unable to trace the contributor of that code to gain their
agreement for the licence change so the code has to be removed.
This commit reverts that contribution. The contribution had no functional
impact so the original way of doing things is still valid. However the
surrounding code has changed significantly so that the exact code as it
was orignally cannot be used. This commit uses the original code as a basis,
but rewrites it to use the PACKET API.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5808)
If a server has been configured to use an ECDSA certificate, we should
allow it regardless of whether the server's own supported groups list
includes the certificate's group.
Fixes#2033
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/5601)
The alpn_selected value in the session should be NULL before we first
populate it if this is a new session. We assert to make sure it is.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5715)
For DTLS/SCTP we were waiting for a dry event during the call to
tls_finish_handshake(). This function just tidies up various internal
things, and after it completes the handshake is over. I can find no good
reason for waiting for a dry event here, and nothing in RFC6083 suggests
to me that we should need to. More importantly though it seems to be
wrong. It is perfectly possible for a peer to send app data/alerts/new
handshake while we are still cleaning up our handshake. If this happens
then we will never get the dry event and so we cannot continue.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5084)
At a couple of points in a DTLS/SCTP handshake we need to wait for a dry
event before continuing. However if an alert has been sent by the peer
then we will never receive that dry event and an infinite loop results.
This commit changes things so that we attempt to read a message if we
are waiting for a dry event but haven't got one yet. This should never
succeed, but any alerts will be processed.
Fixes#4763
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5084)
Fixes regression from #5667.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5701)
Place the session ticket AES and HMAC keys into secure memory.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2351)
Since the public and private DRBG are per thread we don't need one
per ssl object anymore. It could also try to get entropy from a DRBG
that's really from an other thread because the SSL object moved to an
other thread.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/5547)
If the server is configured to allow early data then we check if the PSK
session presented by the client is available in the cache or not. If it
isn't then this may be a replay and we disallow it. If it is then we allow
it and remove the session from the cache. Note: the anti-replay protection
is not used for externally established PSKs.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5644)
We should only update the session cache when we issue a NewSessionTicket.
These are issued automatically after processing a client certificate.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5644)
Conceptually in TLSv1.3 there can be multiple sessions associated with a
single connection. Each NewSessionTicket issued can be considered a
separate session. We can end up issuing multiple NewSessionTickets on a
single connection at the moment (e.g. in a post-handshake auth scenario).
Each of those issued tickets should have the new_session_cb called, it
should go into the session cache separately and it should have a unique
id associated with it (so that they can be found individually in the
cache).
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5644)
Renamed to EVP_PKEY_new_raw_private_key()/EVP_new_raw_public_key() as per
feedback.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5520)
In TLSv1.3 the session is not ready until after the end of the handshake
when we are constructing the NewSessionTicket.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5621)
These functions are similar to SSL_CTX_set_cookie_{generate,verify}_cb,
but used for the application-controlled portion of TLS1.3 stateless
handshake cookies rather than entire DTLSv1 cookies.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5463)
Adds application data into the encrypted session ticket
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3802)
At the core of things is the concept that each extension is only
defined in certain context(s) -- the ClientHello, EncryptedExtensions,
etc., and sometimes only for a specific protocol or protocol range;
we want to enforce that we only parse or generate extensions in the
context(s) for which they are defined. There is some subtlety here,
in that the protocol version in use is not known when generating the
ClientHello (but it is known when the ClientHello extensions are
being parsed!), so the SSL_IS_TLS13() macro must be used with caution.
Nonetheless, by making assertions about whether we are acting in a
server role and whether the current context is (not) a ClientHello,
we can consolidate almost all of the logic for determining whether
an extension is permitted in a given protocol message, whether we
are generating or parsing that message.
The only logic that remains separate relates to generating the ClientHello,
as it depends on an external factor (the maximum permitted TLS version) that
is not defined in the parsing context.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2945)
We also default to SHA256 as per the spec if we do not have an explicit
digest defined.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5554)
They are valid for use in a CertificateRequest message, but we did not
allow it. If a server sent such a message using either of those two
extensions then the handshake would abort.
This corrects that error, but does not add support for actually processing
the extensions. They are simply ignored, and a TODO is inserted to add
support at a later time.
This was found during interoperability testing with btls:
https://gitlab.com/ilari_l/btls
Prompted by these errors I reviewed the complete list of extensions and
compared them with the latest table in draft-24 to confirm there were no
other errors of a similar type. I did not find any.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/5490)
This commit adds SSL_export_keying_material_early() which exports
keying material using early exporter master secret.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5252)
Coverity was complaining because we checked if s->ctx is NULL and then
later on in the function deref s->ctx anyway. In reality if s->ctx is
NULL then this is an internal error.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/5334)
The s_client psk_use_session_cb callback has a comment stating that we
should ignore a key that isn't suitable for TLSv1.3. However we were
actually causing the connection to fail. Changing the return value fixes
the issue.
Also related to this is that the early_data extension was not marked as
TLSv1.3 only which it should be.
Fixes#5202
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/5205)
On the client we calculate the age of the ticket in seconds but the server
may work in ms. Due to rounding errors we could overestimate the age by up
to 1s. It is better to underestimate it. Otherwise, if the RTT is very
short, when the server calculates the age reported by the client it could
be bigger than the age calculated on the server - which should never happen.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5306)
In <= TLSv1.2 a Finished message always comes immediately after a CCS
except in the case of NPN where there is an additional message between
the CCS and Finished. Historically we always calculated the Finished MAC
when we processed the CCS. However to deal with NPN we also calculated it
when we receive the Finished message. Really this should only have been
done if we hand negotiated NPN.
This simplifies the code to only calculate the MAC when we receive the
Finished. In 1.1.1 we need to do it this way anyway because there is no
CCS (except in middlebox compat mode) in TLSv1.3.
Coincidentally, this commit also fixes the fact that no-nextprotoneg does
not currently work in master.
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5285)
Add SSL_verify_client_post_handshake() for servers to initiate PHA
Add SSL_force_post_handshake_auth() for clients that don't have certificates
initially configured, but use a certificate callback.
Update SSL_CTX_set_verify()/SSL_set_verify() mode:
* Add SSL_VERIFY_POST_HANDSHAKE to postpone client authentication until after
the initial handshake.
* Update SSL_VERIFY_CLIENT_ONCE now only sends out one CertRequest regardless
of when the certificate authentication takes place; either initial handshake,
re-negotiation, or post-handshake authentication.
Add 'RequestPostHandshake' and 'RequirePostHandshake' SSL_CONF options that
add the SSL_VERIFY_POST_HANDSHAKE to the 'Request' and 'Require' options
Add support to s_client:
* Enabled automatically when cert is configured
* Can be forced enabled via -force_pha
Add support to s_server:
* Use 'c' to invoke PHA in s_server
* Remove some dead code
Update documentation
Update unit tests:
* Illegal use of PHA extension
* TLSv1.3 certificate tests
DTLS and TLS behave ever-so-slightly differently. So, when DTLS1.3 is
implemented, it's PHA support state machine may need to be different.
Add a TODO and a #error
Update handshake context to deal with PHA.
The handshake context for TLSv1.3 post-handshake auth is up through the
ClientFinish message, plus the CertificateRequest message. Subsequent
Certificate, CertificateVerify, and Finish messages are based on this
handshake context (not the Certificate message per se, but it's included
after the hash). KeyUpdate, NewSessionTicket, and prior Certificate
Request messages are not included in post-handshake authentication.
After the ClientFinished message is processed, save off the digest state
for future post-handshake authentication. When post-handshake auth occurs,
copy over the saved handshake context into the "main" handshake digest.
This effectively discards the any KeyUpdate or NewSessionTicket messages
and any prior post-handshake authentication.
This, of course, assumes that the ID-22 did not mean to include any
previous post-handshake authentication into the new handshake transcript.
This is implied by section 4.4.1 that lists messages only up to the
first ClientFinished.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4964)
If a server receives an unexpected ClientHello then we may or may not
accept it. Make sure all such decisions are made in the state machine
and not in the record layer. This also removes a disparity between the
TLS and the DTLS code. The TLS code was making this decision in the
record layer, while the DTLS code was making it later.
Finally it also solves a problem where a warning alert was being sent
during tls_setup_handshake() and the function was returning a failure
return code. This is problematic because it can be called from a
transition function - which we only allow fatal errors to occur in.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5190)
The new extension is like signature_algorithms, but only for the
signature *on* the certificate we will present to the peer (the
old signature_algorithms extension is still used for signatures that
we *generate*, i.e., those over TLS data structures).
We do not need to generate this extension, since we are the same
implementation as our X.509 stack and can handle the same types
of signatures, but we need to be prepared to receive it, and use the received
information when selecting what certificate to present.
There is a lot of interplay between signature_algorithms_cert and
signature_algorithms, since both affect what certificate we can
use, and thus the resulting signature algorithm used for TLS messages.
So, apply signature_algorithms_cert (if present) as a filter on what
certificates we can consider when choosing a certificate+sigalg
pair.
As part of this addition, we also remove the fallback code that let
keys of type EVP_PKEY_RSA be used to generate RSA-PSS signatures -- the
new rsa_pss_pss_* and rsa_pss_rsae_* signature schemes have pulled
the key type into what is covered by the signature algorithm, so
we should not apply this sort of compatibility workaround.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5068)
Probably this is the CCS between the first and second ClientHellos. It
should be ignored.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4435)
This just adds the various extension functions. More changes will be
required to actually use them.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4435)
Although this is forbidden by all three(!) relevant specifications,
there seem to be multiple server implementations in the wild that
send it. Since we didn't check for unexpected extensions in any
given message type until TLS 1.3 support was added, our previous
behavior was to silently accept these extensions and pass them over
to the custom extension callback (if any). In order to avoid
regression of functionality, relax the check for "extension in
unexpected context" for this specific case, but leave the protocol
enforcment mechanism unchanged for other extensions and in other
extension contexts.
Leave a detailed comment to indicate what is going on.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4463)
